Concepts for the efficient Monte Carlo-based treatment plan optimization in radiotherapy

Abstract

Monte Carlo (MC) dose calculation algorithms are regarded as the gold standard in intensity-modulated radiation therapy (IMRT). Simply adding a MC dose calculation engine to a standard IMRT optimization framework is possible but computationally inefficient. Thus, the optimization would be too time consuming for clinical practice. In this work we developed a hybrid algorithm for the treatment plan optimization that combines the accuracy of MC simulations with the efficiency of less precise dose calculation algorithms. Two methods are introduced that allow a rapid convergence of the iterative optimization algorithm and preserve the efficiency of the MC dose calculation. The performance of the hybrid optimization algorithm is analyzed on different treatment sites. The results are compared against a reference optimization algorithm, which is based on MC simulations in the standard IMRT framework. For this comparison we evaluated several indicators of treatment plan quality, convergence properties, calculation times and efficiency ratios. The efficiency of the optimization could be improved from originally 10-30% to 80-95%. Due to this improvement the calculation times could be reduced to 2-28 minutes, depending on the treatment plan complexity. At the same time, the treatment plan quality could be maintained compared to the reference algorithm.